Momentic
Momentic is an AI testing platform for web, iOS and Android. Teams describe end-to-end tests in plain English, and agents write, run, repair and maintain them as the product changes, locally or in CI.
What is Momentic?
Momentic is an agentic quality platform for end-to-end testing of web, iOS and Android applications, published by Momentic Inc. in San Francisco. Its premise is that a test should describe intent rather than implementation: a test is a plain-English YAML file, versioned in the customer's own repository, with no CSS or XPath selectors to maintain. Three primitives run through the site's examples — act for an action written in natural language, assert for a verification, and module for a reusable block such as a login or a checkout form.
Around that format sit five advertised pillars. The agent first learns the product from documentation, guides, session recordings, tickets and code, and from tools such as Jira, Linear and Figma, building a graph of screens and user flows that it uses to navigate and to spot coverage gaps. Specs then self-heal: when the interface shifts, tests adapt instead of failing. The suite keeps pace with the codebase by watching every commit and pull request. Failures come back actionable, sorted into real bugs, flakiness or environment problems, with session replays, reproduction steps and root-cause analysis. Finally, the whole thing runs continuously inside integration pipelines.
Execution is deliberately portable. The same suite runs on a developer's machine, in any CI able to execute Node.js — GitHub Actions and GitLab are documented — or on Chromium browsers, Android emulators and iOS simulators hosted by Momentic. A command-line interface drives it all, with momentic run, momentic ai classify, momentic ai heal and momentic ai explore, while an MCP server lets coding agents such as Claude Code or Cursor operate the platform directly. A separate agent called Mo, presented as an AI QA that bug-bashes an application from a single objective sentence, is announced on a waiting list rather than generally available.
The scale claimed on the site is industrial: 77,813 tests created, more than 70 million test runs executed, 8.9 million auto-heals, 117,010 bugs caught and 80,290 pull requests verified, at a stated 96% signal-to-noise ratio. Notion, Quora, Retool, Webflow, Xero and GPTZero appear as named customers. Enterprise buyers are offered SOC 2 Type 2, a 99.99% uptime SLA, SAML SSO, role-based access control and audit logs.
What it does
- Write end-to-end tests in plain English, stored as versioned YAML files in your own repository
- Run suites on local machines, in CI, or on hosted browsers, Android emulators and iOS simulators
- Auto-heal tests when the interface changes, instead of rewriting broken selectors
- Classify and triage failures as real bugs, flakiness or environment issues
- Trace a failure to its root cause with session replays and reproduction steps
- Generate new tests from pull requests and diffs, and explore the app to close coverage gaps
- Verify every commit, pull request and deployment, reporting to Slack, email or the Momentic app
When to use Momentic / When not to
A quick filter to help you decide if Momentic is the right fit.
When to use Momentic
- Engineering teams shipping daily, whose manual QA can no longer keep up with the release cadence
- Teams dragging a flaky end-to-end suite that breaks every time the interface moves
- Product teams needing one workflow across Chromium browsers, iOS simulators and Android emulators
- AI product teams validating non-deterministic model outputs, as Quora does for Poe.com responses
- Startups and solo developers who want a permanent free tier before committing any budget
When not to use Momentic
- Non-technical users: every workflow assumes a code repository and a CI pipeline
- Teams looking for unit tests or isolated API testing, outside the advertised end-to-end scope
- Buyers who require guaranteed European hosting or a published data processing agreement
- Organisations needing an interface or documentation in a language other than English
- Users located in countries under United States embargo, explicitly excluded by the terms of use
How to use Momentic
A typical end-to-end flow, from setup to results.
- Create a free account at app.momentic.ai/signup: no credit card is required, and Google SSO works on every plan
- Point Momentic at your application and let the agent learn the product from your docs, tickets and codebase
- Write a first test in the web spec editor, or straight from the terminal with the CLI
- Describe the steps in plain English inside a YAML file, with a start URL and a short description
- Factor recurring sequences such as sign-in or form filling into reusable .module.yaml blocks
- Run it locally with momentic run, adding --parallel auto to size the worker pool to your CPU count
- Wire the suite into CI so it runs on every push from GitHub Actions, GitLab or any Node.js provider
- Provision disposable inboxes with email.create() and numbers with sms.lease() for sign-up and OTP flows
- Review results in the hosted Run Viewer, or locally with momentic results view without uploading anything
- Route reports to Slack, email or the Momentic app, and let the agent auto-heal or classify what failed
Pros & Cons
Pros
- Tests are written in plain English and stored as YAML in your own repository, so they stay readable and versioned
- Maintenance is largely absorbed by auto-healing, with 8.9 million auto-heals claimed on the site
- One workflow covers web, iOS and Android, running locally, in CI or on hosted infrastructure
- Usage-based pricing with no per-seat charge: the entire team can be invited at no extra cost
- The free plan is permanent, needs no credit card and stops hard, so no surprise invoice is possible
- Published customer outcomes are specific: Quora cut daily execution from seven hours to thirty minutes, Retool reports an eightfold release cadence, GPTZero an 89% drop in escaped defects
- SOC 2 Type 2 certification and a 99.99% uptime SLA are available for enterprise requirements
Cons
- No training opt-out is documented, while the terms explicitly allow third-party AI providers to train on inputs and outputs
- The licence taken over customer data is unusually broad: perpetual, irrevocable, sublicensable and expandable at the vendor's discretion
- No data processing agreement is published, and the Trust Center returns no readable content without a browser
- No European hosting region is announced, and no Article 27 EU representative is named
- Credits never roll over, and full run results are deleted after 30 days outside Enterprise
- SAML SSO, SCIM, audit logs and the uptime SLA are Enterprise-only, and Enterprise pricing is not public
- Mo, the AI QA agent promoted at the top of every page, is still a waiting list rather than a product
Pricing & Plans
A permanent free plan is available at no cost and without a credit card, providing 2,000 credits per month, roughly 200 test runs. The lowest paid entry point is the Pay-as-you-go plan at USD 125 per month, which includes 10,000 credits and bills additional usage at USD 0.01875 per credit. Enterprise pricing is test-based and is not published.
- 2
- 000 credits per month (about 200 runs)
- all core features and AI test authoring
- 30-day results retention
- one concurrent mobile device
- 30 minutes of mobile emulator time per month
- Google SSO
- no credit card required
- everything in Free
- 10
- 000 credits per month (about 1
- 000 runs)
- no hard cap
- top-ups of 10
- 000 credits for USD 125
- overage at USD 0.01875 per credit
- everything in Pay-as-you-go
- test-based pricing
- SAML SSO
- SCIM provisioning
- audit log
- domain claiming
- unlimited mobile session length
- higher API rate limits
Data, GDPR & hosting
A consolidated view of how Momentic handles your data.
GDPR overview
The privacy policy, effective 8 May 2026, devotes a full section to UK and EEA residents. It publishes a table of legal bases: contract performance and legitimate interest for delivering and improving the service, contract for support and payment, legitimate interest for marketing, consent for cookies and pixels. Momentic designates itself controller for data collected through the website and processor for data handled through the service. Cross-border transfers rely on adequacy decisions, European Commission standard contractual clauses or binding corporate rules. Access, rectification, erasure and objection rights are exercised through privacy@momentic.ai or the Your Privacy Choices page, with an appeal procedure and a right to complain to the local supervisory authority. Two gaps stand out: no Article 27 EU representative is named, and no data processing agreement is published or even mentioned.
Who owns the data?
Under section 4.1 of the terms of use, the customer owns and retains every right, including intellectual property rights, in the data uploaded to or generated through the service, and inputs sent to the AI features remain customer data. In exchange, section 4.2 grants Momentic a worldwide, royalty-free, transferable, sublicensable, perpetual and irrevocable licence to process that data in order to run, support and improve the service, improve other Momentic products, conduct market research and product development, and for any other lawful purpose. Usage data derived from that activity belongs to Momentic alone under section 6.2, with no access right for the customer. Momentic performs no backup and declines liability for data loss.
Reuse rights
The customer keeps ownership of its inputs and may reuse them freely, without asking permission. Outputs are generated in response to those inputs, and the terms warn that similar inputs may produce the same output for other users, so no exclusivity is guaranteed. Section 6.3.4 places the burden of vetting accuracy, legality and third-party rights on the customer before any output is used. On the vendor side the licence is deliberately broad and may be widened unilaterally where the law does not prohibit it. Section 6.3.5 goes further: third-party providers of the underlying artificial intelligence may access inputs and outputs and use them to train, improve and develop their own technologies. The providers named in the privacy policy are Anthropic, OpenAI, X.AI and Google. No opt-out from that training is documented anywhere on the site.
Data retention & training
Hosting summary
Momentic publishes no dedicated hosting page, and its Trust Center is a JavaScript-rendered application that returns no readable content, so the only authoritative statement available is in the privacy policy. That document states that data may be transferred to and stored in countries other than the one where it was collected, expressly including the United States and Canada, and more broadly in destinations outside the EEA and the United Kingdom. Such transfers rely on European Commission adequacy decisions, standard contractual clauses or binding corporate rules. The publisher is established at 177 Townsend Street in San Francisco, and the domain resolves to an anycast address operated by Amazon Web Services in the United States, which is an infrastructure signal rather than a vendor commitment. No European hosting region is announced. Test execution is a separate question: suites can run on the customer's own machines or CI, in which case application data never reaches Momentic's infrastructure, or on the hosted browsers, emulators and simulators the vendor provides.
Where Momentic works
Country-level availability.
Not available in
Things to keep in mind
Risks and trade-offs to weigh before adopting Momentic.
- Inputs and outputs may be used by third-party AI providers to train their own models, and no opt-out is documented
- The licence over customer data is perpetual and irrevocable, and the vendor reserves the right to widen its use unilaterally
- Question-and-answer interactions are stored indefinitely alongside the account and are removed only on request
- Delegating verification to agents can erode a team's own grasp of its critical journeys, since the specs become the only place that knowledge is written down
- Credits expire at the end of each cycle and overage costs half as much again as a top-up, so unplanned usage is penalised twice
- When hosted infrastructure is chosen, application and test data flow through browsers and emulators operated by the vendor
- The terms impose individual AAA arbitration in San Francisco and the exclusive jurisdiction of San Francisco federal courts
Setup & Integrations
Technical difficulty
Moderate, and clearly aimed at engineers. Signing up is instant and free, and writing a test needs no framework knowledge: steps are plain English, with no selectors to maintain. The technical floor lies elsewhere. Tests are YAML files living in a repository, the CLI has to be installed and run, and the real payoff arrives once the suite is wired into a CI pipeline. Quickstarts cover web, iOS, Android and the CLI, an IDE lint validates test files as they are written, and no infrastructure work is needed if hosted browsers and emulators are used.
Deployment
Integrations
Supported languages
Behind Momentic
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Momentic.
Frequently asked questions
How does Momentic charge for testing?
Is there a free plan, and does it require a credit card?
What happens when the credits run out?
Which platforms and environments are supported?
Does Momentic provide an API?
Is customer data used to train AI models?
Who owns the data and the test results?
How long are test results kept?
What security and compliance guarantees are offered?
Should you pick Momentic?
Momentic is a credible bet on a simple idea: an end-to-end test should express user intent rather than DOM structure, and the maintenance it generates should fall to agents rather than engineers. Two years after its domain first appeared online, the company ships across web, iOS and Android, has raised USD 18.7 million over two announced rounds, holds SOC 2 Type 2 certification and publishes named reference customers including Notion, Quora, Retool and Webflow. The usage figures it advertises, more than 70 million test runs and 8.9 million auto-heals, describe a product genuinely in production rather than a demo.
The commercial model is unusually legible. Billing follows credits rather than seats, so inviting the whole team costs nothing, and the free tier is a permanent plan with a hard stop rather than a countdown to an invoice. Evaluation is therefore cheap and low-risk.
The reservations concentrate on data governance, and they deserve reading before signing. The licence taken over customer data is perpetual, irrevocable, sublicensable and expandable at the vendor's discretion; the usage data derived from that activity belongs to Momentic; and the terms explicitly permit third-party AI providers to train on inputs and outputs, with no opt-out documented anywhere. The privacy policy asserts the opposite restriction, a contradiction a legal reader should resolve directly with the vendor. No data processing agreement is published, no European hosting is offered and no Article 27 representative is named, all of which matter to a European buyer.
One caveat on scope, finally: Mo, the AI QA agent promoted at the top of every page, is still a waiting list. What ships today is the test automation layer, and it deserves to be judged on that.
- Choosing a selection results in a full page refresh.
- Opens in a new window.